The discussion section occupies a central role in research articles because ideally such sections crystallize how the paper at hand has contributed to a research field. Annesley (2010, p. 1671) elucidates the significance of the discussion section by characterizing the reading habits of scientists. When reading a paper, they first tend to look at the abstract to get an overview of the topic and the purported findings. If the topic appears to be of interest, they move to the discussion section. If it is neither stimulating nor convincing about the meaning and importance of the findings, it does not really matter how the experiments were performed or what results were reported in the findings section.

So far, the features of discussion sections have mainly been analysed in language studies. Many of these investigations draw on genre analysis, with the intent to specify the schematic structures of discussion sections. As a result, diverse structural constituents referred to as moves and steps have been identified. Moves are linguistic and rhetorical constructs, and they include, for example, the comparison of findings with the observations of prior investigations and the specification of the theoretical contributions of the study. Steps are subconstituents of moves; examples of steps include the explanation of a surprising finding and the evaluation of the significance of an observation (Al-Shujairi et al., 2019). Since the 1980s, genre studies have offered a detailed picture of the structural features of discussion sections in diverse fields such as Applied Language Science, Computer Research, Psychology and Sociology. On the other hand, the studies scrutinizing moves and steps are limited in that the findings offer only a surface picture of the content of discussion sections. The results just provide illustrative examples of how the content of research findings appears within structural constituents.

The present investigation pioneers in the domain of information behaviour research by examining the role of discussion sections in a health information seeking studies. The findings deepen our understanding about how researchers examining the issues of health information seeking crystallize their contributions and reflect their significance. This domain was chosen for the study because it represents an established subdomain of information behaviour research (Given, Case and Willson, 2023, pp. 81-85). In general, health information seeking deals with the ways in which people identify, select and access information about their health, health promotion activities, risks to one’s health, and illness (Lambert and Loiselle, 2007; Mirzaei et al., 2021). Given the broad repertoire of the topics of health information seeking, it may be expected that the discussion sections of research articles offer a fertile ground for the analysis of the ways in which scholars reflect their contributions to this particular domain. The present investigation draws on the frameworks of genre-based studies referred above. However, different from them, the focus is placed on the constituents´ content in discussion sections. The findings are also relevant from the perspective of scholarly communication because the writing of research articles offers a major way to publish the results of scientific work.

The rest of the article is organized as follows. The literature review characterises further the nature of discussion sections and depicts the genre analytic approach to the topic. Thereafter, the research framework of the present investigation is specified, followed by the formulation of the research questions and the description of the empirical data and its analysis. Next, the research findings are reported, and their significance is reflected.

The literary review suggests that discussion sections form a significant part of research articles because these sections crystallize the key findings of the study, explain their significance, reflect the study´s contribution to a research domain, and identify the limitations of the investigation. The literature review also indicated that genre-based studies have concentrated on the structural constituents (moves and steps) of discussion sections, particularly the order in which they appear in the texts. While the present study draws on the above findings while identifying the relevant constituents of discussion sections, it also introduces a new viewpoint by concentrating on how researchers fill the containers labelled as constituents with particular content pertinent to health information seeking.

The research framework of the present study was developed by making use of the models for discussion sections reviewed above. Due to their generic nature and applicability across diverse disciplines, the constituents identified by Peacock (2002) and Al-Shujairi et al. (2019) were particularly important. For the empirical analysis of the content of discussion sections in articles examining health information seeking, the following constituents identified by Peacock (2002) appeared to be relevant:

In addition, from the model developed by Al-Shujairi et al. (2019), the following constituents were relevant for the empirical analysis:

Differently from the models reviewed above, the present study will not examine the order in which the constituents of the discussion section appear in the texts. As the study focuses on the constituents´ content, the order in which the content is presented in the discussion sections is of secondary importance. The preliminary analysis of the empirical material of the present investigation also revealed that the differentiation between moves and steps would not bring additional value to the analysis focusing on the constituents´ content. It is sufficient to operate at a more general level by approaching the linguistic and rhetorical units pertinent to the content of discussion sections in terms of constituents. The research framework was substantiated by including additional constituents identified inductively from the empirical data of the present study, that is, the sample of 100 articles examining health information seeking. These constituents deal with the ways in which researchers interpret their findings and compare them to the results of prior investigations. More specifically, the following constituents are relevant in this regard:

In addition, the following constituents depicting the ways in which researchers reflect their contributions as a whole were identified inductively from the research material:

The research framework of the present study was finalised by taking the constituents identified by Peacock (2002) and Al-Shujairi et al. (2019) as a point of departure. However, some of the constituents were renamed more informatively for the needs of the present investigation. For example, the constituent of “Expected or unexpected outcome (comment on whether the result is expected or not)’ (Peacock, 2002) was renamed as Interpreting individual findings, while the constituent of Background information (Al-Shujairi et al., 2019) is referred to as Offering background information about the study and its context. The initial framework was substantiated by adding the constituents inductively identified from the empirical material. Although the specific order of the constituents in discussion sections will not be examined, the constituents are divided into three groups reflecting the major parts of discussion sections. Dudley-Evans (1994) identified three macro-level parts of the discussion section, that is, introduction, evaluation, and conclusion. For the purposes of the present investigation, these parts were correspondingly renamed as Introduction, Interpretation of findings, and Specification of contributions and limitations. The research framework is specified in Table 1. The illustrative examples of the constituents are taken from the empirical material of the present investigation.

While discussing their findings, the researcher can make use diverse ways to reflect the contributions of an article. In the introductory part of the discussion section, they may offer background information about the goals of the study, as well as summarise the key findings. Thereafter, in the middle part of the section, the research findings are interpreted. To this end, the findings may be compared to the observations of prior studies by depicting similarities and differences. A more analytical way to interpret the findings is to explain why they are consistent or inconsistent with prior results. In the concluding part, the researcher moves from the interpretation of individual findings to the specification of their significance as a whole. This can be done by showing how the results offer an empirical, theoretical, methodological and/or practical contribution. In the final part, based on their findings, the researcher can also identify gaps in prior research, assess the limitations of the study, as well as suggest topics for future research.

To examine how researchers in the domain of health information seeking crystallize their contributions and reflect their significance, the present study draws on the research framework presented in Table 1. More specifically, the present investigation seeks answers to the following questions.

RQ3. In the above sections, how do researchers specify their contributions to health information seeking studies as a whole?

The empirical research material was identified in April 2024 by searching Library and Information Science Abstracts (Proquest). This database was chosen because it extensively covers studies on information-seeking behaviour occurring in diverse contexts. Using the search term health information seeking, the database searches were directed to the abstracts of the articles on the above topic. For clarity and to avoid terminological ambiguity, the search term health information searching was not used. This is because researchers sometimes use it interchangeably with the term health information seeking, without specifying how health information searching differs from health information seeking (Bachl et al., 2024). In the searches, three criteria were employed in order to identify pertinent material. First, the topic´s relevance was used as criterion. Second, the searches were limited to peer-reviewed articles, thus leaving out material of other types. Third, only articles published in English were included in order to ensure that the terminology used in the articles is comparable. By the above criteria, a sample of the 100 first articles of the search list was downloaded for a closer review. The sample appeared to be sufficient for the needs of a qualitative study because the material became saturated. It became evident that the inclusion of additional articles would not have essentially changed the qualitative picture of the discussion sections of articles examining the issues of health information seeking.

The 100 articles were published in 27 diverse journals within the period of 2009-2024. Examples of the titles of the articles include “Laypeople’s source selection in online health information-seeking process” and “The association between health information seeking on the Internet and physician visits”. The list of articles is presented in Appendix 1. The most frequent publication forums of the articles were Journal of Medical Internet Research (30 articles), Journal of Documentation (13 articles), Aslib Journal of Information Management (8 articles), Library & Information Science Research (6 articles) and Online Information Review (5 articles). The rest of the articles were scattered amongst a number of journals, for example, Information Research, Journal of the Association for Information Science and Technology, and Health Communication. Of the 100 articles, 71 were based on quantitative research (e.g., questionnaire surveys), while 12 articles were qualitative investigations based on the analysis of interview data. Moreover, 12 articles were systematic literature reviews, and 5 articles drew on the mixed methods approach. The length of the discussion sections varied between 569 and 3626 words. On average, a discussion section contained 1468 words. There was some variation in the ways in which the authors employed guiding subtitles in the discussion sections in order to orient the readers. For example, in the Journal of Medical Internet Research, most discussion sections were initiated with a text paragraph placed under the subtitle of Principal findings. In addition, the authors employed various subtitles such as Comparison with prior work, and Strengths and limitations. It is probable that subtitles such as these were used simply because the instructions for authors published by Journal of Medical Internet Research explicitly require that ‘the following format ("IMRD Format") must be used for the paper’ and that Discussion is structured using ‘e.g., the subheadings "Principal Results", "Limitations", "Comparison with Prior Work", "Conclusions’ (https://www.jmir.org/author-information/instructions-for-authors). In other journals, for example, the Journal of Documentation, and Library & Information Science Research, there were no guidelines for the authors about how to structure the discussion section. Therefore, it is understandable that in these journals, the use of subtitles varied from an article to another, while many discussion sections lacked subtitles entirely.

Following the advice of Basturkmen (2012, p. 136), the downloaded articles were first read in their entirety. This was done to avoid a partial understanding of the discussion sections. Thereafter, the discussion sections were read twice to obtain a preliminary picture of their content. The study was continued by conducting a preliminary coding of the material by making use of the constituents identified by Peacock (2002) and Al-Shujairi et al. (2019). The material was coded by the present researcher; there were no other coders. As noted above, the coding scheme was substantiated by adding constituents inductively identified from the research material, for example, ‘Comparison with prior studies: explaining similarities’, and ‘Offering an empirical contribution’. In the second phase of the coding, all relevant text portions (sentences and text paragraphs) were equipped by codes indicative of the constituents specified in Table 1 so that an individual code was used one or more times within a discussion section. For example, every single research finding interpreted by the author of an article was coded separately, as well as every theoretical implication specified by the researcher. However, sentences in which the authors returned to individual research results already depicted in the findings section - without interpreting them in any way - were excluded from because these text portions did not offer new content to the discussion section. The coding was refined and checked several times by the present author until it was concluded that the codes appropriately cover the whole research material and that there are no anomalies. More specifically, to strengthen the reliability of the coding, the initial coding was refined by repeated reading of the data. During this process, the methodological recommendation of Miles and Huberman (1994, p. 65) was followed: the careful checking of the codes is a useful method for the lone researcher if the code-recode consistencies reach at least 90%. Following this advice, the coding was refined until it was found that the codes appropriately describe the data and that there are no anomalies.

The coded material was analysed both quantitatively and qualitatively by the present author. First, to obtain an indicative quantitative picture of the content of discussion sections, the distribution of the codes assigned to the constituents was calculated by means of descriptive statistics. The analysis was refined by means of qualitative content analysis (Zhang and Wildemuth, 2016). More specifically, attention was directed to the variation in the constituents´ content. The analysis of the variation was initiated by the identification of similarities and differences in the ways in which the authors depicted such content per constituent, for example, while summarizing the key findings and reflecting the empirical contribution. In the next phase of the analysis, the focus was placed on the comparison of such similarities and differences per constituent. This enabled the scrutiny of the various ways in which the authors, for example, explained why their findings are consistent with earlier studies on the topic or how the research results contribute to the development of models for health information seeking.

Drawing on the ideas of genre analysis, the present study elaborated the picture of the ways in which researchers in a particular domain reflect their contributions in the discussion sections of scholarly articles. While genre-based studies have offered a detailed view on the structure and constituents of discussion section, the present investigation demonstrated how researchers fill the structural constituents with content in the domain of health information seeking. More specifically, the present study adds knowledge about how researchers in the above domain realise the norms of scholarly writing by generating content for the “containers” represented by diverse constituents in discussion sections. The main findings of are summarized in Table 3.

The present sought answers to three research questions. First, it was asked how do researchers offer background information and summarize their findings in the discussion sections? In this regard, as Table 3 suggests, the research results are unsurprising. Researchers write the introductory part to remind readers about the research goal of the study and to direct their attention to the key findings of the investigation. Following the academic convention, the introductory part is usually commonly kept concise because it represents a necessary first step on the way to the core content of the discussion.

The second research question focused on the ways in which researchers interpret the individual key findings. This occurs in the middle part of the discussion section. First, while interpreting individual research results, the writers emphasized their novelty or surprise value; a phenomenon or relation is identified for the first time in health information seeking studies. Second, an individual finding was interpreted by drawing on descriptive comparison, that is, depicting similarities and differences between one´s finding and prior research results. The popularity of similarity description suggests that researchers prefer a conservative approach to the interpretation of their findings. The emphasis is placed on their confirmation in light of prior observations, rather than exposing one´s finding to opposing or contradictory evidence offered by earlier investigations. Third, an individual finding was interpreted using explanatory comparison. To this end, the researcher explicated why his or her finding is consistent or inconsistent with prior observations. Explanatory comparison was employed quite seldom, probably because this approach is cognitively more demanding than descriptive comparison.

Finally, the third research question dealt with the concluding part of the discussion section, that is, the ways in which researchers specify their contributions to health information seeking studies as a whole. In this regard, four major strategies were identified. First, the writers evaluated how their findings refined or complemented the empirical picture of health information seeking by identifying new phenomena or statistically significant relationships between variables. Second, researchers specified how their findings resulted in the elaboration of models for health information seeking or offered a new methodological viewpoint to health information seeking studies. The specification of the contributions to model development and methodology were rare, thus suggesting that they are cognitively demanding tasks, requiring that the researcher is able to rise above his or her individual research results. Third, the writers reflected the practical implications of their findings. This is not necessarily easier than the pondering of the theoretical implications because practical suggestions are more convincing if the writer is able to demonstrate how his or her findings can be applied to the design of a particular information service, for example. Fourth, the contributions were specified by identifying the limitations of the study and explaining how they could be overcome in future investigations. Closely related, the identification of gaps in prior studies served the same end because the lack of investigations in a domain is indicative of a potentially relevant research topic.

Given that the present study pioneers in the analysis of discussion sections dealing with health information seeking, there is a paucity of prior investigations offering opportunities for comparative notions. The findings support Hyland´s (2008, p. 3) assumption about researchers´ ways to draw on the ‘persuasive practices of their discipline, encoding ideas, employing warrants, and framing arguments in ways that their potential audience will find most convincing’. The discussion sections examined in the present study indicate that the persuasive practices include, for example, the highlighting of the novelty value of the findings, thus convincing the readers about the relevance of research approach. The findings also lend support to the conclusion drawn by Arsyad et al. (2020, p. 294) and Avidan, Ioannidis and Mashour (2019, p. 414) They claimed that writers tend to refer to the findings of prior studies in order to demonstrate that their own investigation is an integral part of a relevant research tradition and that their findings constructively add knowledge to it. This preference was clearly evident, as the authors favoured the seeking of confirmatory support rather than contrary evidence while comparing their findings with earlier investigations. While reflecting the contributory role of discussion sections, Kearney (2017, p. 290) likened an informative discussion section to a “road map” which is capable of demonstrating for the readers ‘where they are after this study, in comparison to where they were before’. The discussion sections examined in the present study seldom achieved this goal. This is mainly due to the paucity of reflection focusing on the theoretical contributions of a study. Investigations based on the systematic review of health information seeking studies fared better in this regard because they disclosed gaps in research and suggested topics of future research.

The findings of the present study, as well as the comparative notions have theoretical implications for the analysis of the discussion sections. The models developed in the domain of genre analysis offer a firm basis for the identification of the constituents of discussion sections, as well as the scrutiny of the order in which the constituents appear in the texts. As the genre analytic approaches are focused on the structure of the scholarly articles, there is a need to refine the analysis of the content of the constituents. To this end, the Argumentative Zoning (AZ) theory developed by Teufel (1999) offers useful conceptual and methodological tools. AZ is an analysis of the argumentative and rhetorical structure of a scientific paper (Teufel, Siddhartan and Batchelor, 2009, p. 1493. One of the strengths of AZ is that it enables a sentence-by-sentence classification with mutually exclusive categories from the annotation scheme incorporating categories such as Aim, Background, Basis, Contrast, Own, and Textual. For example, the category of Aim points out the paper’s main knowledge claim, a rhetorical move which may be repeated in the introductory section and Discussion, while the category Textual explains the physical location of information, e.g., by giving a section overview or presenting a summary of a subsection.

The findings also have practical implications for the writers of discussions sections. They could be enhanced by directing more attention to the reflection of contrasting evidence in order to test the relevance of interpretations. Moreover, there is a need to employ more deeply analytical strategies based on the explanatory comparison. This approach is cognitively more demanding than description and comparison; however, it can enable digging deeper, possibly leading to the identification of new connections between one´s findings and the observations of prior research on the topic. There is also a need to demonstrate in more detail how one´s findings would contribute to the conceptual and theoretical development in a domain. This would serve the ends of cumulating knowledge in a field because the findings would be put into a broader context and thereby be used more effectively in future investigations on the topic.

Basturkmen (2012, p. 143) has made useful suggestions for future studies on discussion sections. Interviews with the writers could offer useful information about the ways in which researchers compose the discussion sections. Interviews would be most effective if they are directed to a recently written discussion section because the authors could be able to recall better the writing process. Examples of relevant interview questions include, why was the discussion section composed in a certain way, how was the content of the text drafted and finalized, and how did the author draw on diverse strategies to reflect his or her contributions? On the other hand, the writers may find such interviews demanding because much of the knowledge used by the researcher is likely be tacit, thus subject to intuitive thinking and insights that arose during the writing process. Therefore, it is possible that the authors cannot recall and articulate exactly how they filled the various constituents of the discussion with particular content. Nevertheless, the interview data would offer valuable material for the elaboration of the picture of the scholarly writing process and academic conventions behind it.

Discussion sections occupy a significant role when researchers reflect their empirical, theoretical, methodological and practical contributions. The present investigation pioneered by demonstrating how researchers in the domain of health information seeking interpret their findings. The results highlight that in the above domain, the authors tend to prefer a conservative approach by seeking confirmatory support from prior studies. As the present investigation focused on a sample of 100 discussion sections, the results do not offer a generalizable picture of the content of such sections in the domain of health information seeking. Thus there is a need to expand the analysis by examining how researchers generate and justify their contributions in diverse subdomains of information behaviour research. Another topic of future research is a more detailed analysis of the content of individual constituents, for example, the articulation of theoretical and practical implications. Examples of the topics of future studies also include the comparison of discussion and conclusion sections. The comparative approach is meaningful because conclusion sections may have unique elements. Lamanauskas (2021, pp. 7-8) referred to them by suggesting that ideally, conclusion sections are able to express the ‘purification of the gist of the research carried out’. Similar to an effective final paragraph of a discussion section, Conclusion can offer a “take-home message”, that is, a crystallised description of the main contribution of an investigation.

Authors contributing to Information Research agree to publish their articles under a Creative Commons CC BY-NC 4.0 license, which gives third parties the right to copy and redistribute the material in any medium or format. It also gives third parties the right to remix, transform and build upon the material for any purpose, except commercial, on the condition that clear acknowledgment is given to the author(s) of the work, that a link to the license is provided and that it is made clear if changes have been made to the work. This must be done in a reasonable manner, and must not imply that the licensor endorses the use of the work by third parties. The author(s) retain copyright to the work. You can also read more at: https://publicera.kb.se/ir/openaccess